A diverse selection of 1000 empirical time series, along with results of an hctsa feature extraction, using v1.06 of hctsa and Matlab 2019b, computed on a server at The University of Sydney.The results of the computation are in the hctsa file, HCTSA_Empirical1000.mat for use in Matlab using v1.06 of hctsa.The same data is also provided in .csv format for the hctsa_datamatrix.csv (results of feature computation), with information about rows (time series) in hctsa_timeseries-info.csv, information about columns (features) in hctsa_features.csv (and corresponding hctsa code used to compute each feature in hctsa_masterfeatures.csv), and the data of individual time series (each line a time series, for time series described in hctsa_timeseries-info.csv) is in hctsa_timeseries-data.csv. These .csv files were produced by running >>OutputToCSV(HCTSA_Empirical1000.mat,true,true); in hctsa.The input file, INP_Empirical1000.mat, is for use with hctsa, and contains the time-series data and metadata for the 1000 time series. For example, massive feature extraction from these data on the user's machine, using hctsa, can proceed as>> TS_Init('INP_Empirical1000.mat');Some visualizations of the dataset are in CarpetPlot.png (first 1000 samples of all time series as a carpet (color) plot) and 150TS-250samples.png (conventional time-series plots of the first 250 samples of a sample of 150 time series from the dataset). More visualizations can be performed by the user using TS_PlotTimeSeries from the hctsa package.See links in references for more comprehensive documentation for performing methodological comparison using this dataset, and on how to download and use v1.06 of hctsa.

Dataset

This dataset was created by Tan Phan

Contents

Dataset

This dataset was created by Saurav Anand

Contents

Multivariate Time-Series (MTS) are ubiquitous, and are generated in areas as disparate as sensor recordings in aerospace systems, music and video streams, medical monitoring, and financial systems. Domain experts are often interested in searching for interesting multivariate patterns from these MTS databases which can contain up to several gigabytes of data. Surprisingly, research on MTS search is very limited. Most existing work only supports queries with the same length of data, or queries on a fixed set of variables. In this paper, we propose an efficient and flexible subsequence search framework for massive MTS databases, that, for the first time, enables querying on any subset of variables with arbitrary time delays between them. We propose two provably correct algorithms to solve this problem — (1) an R-tree Based Search (RBS) which uses Minimum Bounding Rectangles (MBR) to organize the subsequences, and (2) a List Based Search (LBS) algorithm which uses sorted lists for indexing. We demonstrate the performance of these algorithms using two large MTS databases from the aviation domain, each containing several millions of observations. Both these tests show that our algorithms have very high prune rates (>95%) thus needing actual disk access for only less than 5% of the observations. To the best of our knowledge, this is the first flexible MTS search algorithm capable of subsequence search on any subset of variables. Moreover, MTS subsequence search has never been attempted on datasets of the size we have used in this paper.

Load, wind and solar, prices in hourly resolution. This data package contains different kinds of timeseries data relevant for power system modelling, namely electricity prices, electricity consumption (load) as well as wind and solar power generation and capacities. The data is aggregated either by country, control area or bidding zone. Geographical coverage includes the EU and some neighbouring countries. All variables are provided in hourly resolution. Where original data is available in higher resolution (half-hourly or quarter-hourly), it is provided in separate files. This package version only contains data provided by TSOs and power exchanges via ENTSO-E Transparency, covering the period 2015-mid 2020. See previous versions for historical data from a broader range of sources. All data processing is conducted in Python/pandas and has been documented in the Jupyter notebooks linked below.

The sales data for the first two products (P1 and P2) are weekly and data was collected until November 10, 2019. Products P3 and P4 are daily and might be related. For product P4, the company has provided potential explanatory variables X1 (price) and X2 (weather forecast of temperature in °C) that may be helpful for forecasting these two products. The sales data for products P3 and P4 was collected until November 24, 2019. Data for product P5 is weekly and was collected until August 30, 2019.

Visualization - https://public.tableau.com/views/ProductSales_16457072047730/Dashboard1?:language=en-US&publish=yes&:display_count=n&:origin=viz_share_link

This dataset contains daily rainfall measurements (in millimeters) for the year 2022. The data spans from January 1, 2022, to July 3, 2022, covering a total of 184 days. The dataset can be used for various machine learning tasks, such as time series forecasting, pattern recognition, or anomaly detection related to rainfall patterns.

Column Descriptors:

date (date): Description: The date of the rainfall measurement in the format YYYY-MM-DD. Example: 2022-01-01 rainfall (float): Description: The amount of rainfall recorded on the corresponding date, measured in millimeters (mm). Example: 12.5 Range: The rainfall values range from 0.0 mm (no rainfall) to 22.4 mm (the maximum recorded value in the dataset). Missing values: There are no missing values in this column.

The Controlled Anomalies Time Series (CATS) Dataset consists of commands, external stimuli, and telemetry readings of a simulated complex dynamical system with 200 injected anomalies.

The CATS Dataset exhibits a set of desirable properties that make it very suitable for benchmarking Anomaly Detection Algorithms in Multivariate Time Series [1]:

Multivariate (17 variables) including sensors reading and control signals. It simulates the operational behaviour of an arbitrary complex system including:

4 Deliberate Actuations / Control Commands sent by a simulated operator / controller, for instance, commands of an operator to turn ON/OFF some equipment.

3 Environmental Stimuli / External Forces acting on the system and affecting its behaviour, for instance, the wind affecting the orientation of a large ground antenna.

10 Telemetry Readings representing the observable states of the complex system by means of sensors, for instance, a position, a temperature, a pressure, a voltage, current, humidity, velocity, acceleration, etc.

5 million timestamps. Sensors readings are at 1Hz sampling frequency.

1 million nominal observations (the first 1 million datapoints). This is suitable to start learning the "normal" behaviour.

4 million observations that include both nominal and anomalous segments. This is suitable to evaluate both semi-supervised approaches (novelty detection) as well as unsupervised approaches (outlier detection).

200 anomalous segments. One anomalous segment may contain several successive anomalous observations / timestamps. Only the last 4 million observations contain anomalous segments.

Different types of anomalies to understand what anomaly types can be detected by different approaches. The categories are available in the dataset and in the metadata.

Fine control over ground truth. As this is a simulated system with deliberate anomaly injection, the start and end time of the anomalous behaviour is known very precisely. In contrast to real world datasets, there is no risk that the ground truth contains mislabelled segments which is often the case for real data.

Suitable for root cause analysis. In addition to the anomaly category, the time series channel in which the anomaly first developed itself is recorded and made available as part of the metadata. This can be useful to evaluate the performance of algorithm to trace back anomalies to the right root cause channel.

Affected channels. In addition to the knowledge of the root cause channel in which the anomaly first developed itself, we provide information of channels possibly affected by the anomaly. This can also be useful to evaluate the explainability of anomaly detection systems which may point out to the anomalous channels (root cause and affected).

Obvious anomalies. The simulated anomalies have been designed to be "easy" to be detected for human eyes (i.e., there are very large spikes or oscillations), hence also detectable for most algorithms. It makes this synthetic dataset useful for screening tasks (i.e., to eliminate algorithms that are not capable to detect those obvious anomalies). However, during our initial experiments, the dataset turned out to be challenging enough even for state-of-the-art anomaly detection approaches, making it suitable also for regular benchmark studies.

Context provided. Some variables can only be considered anomalous in relation to other behaviours. A typical example consists of a light and switch pair. The light being either on or off is nominal, the same goes for the switch, but having the switch on and the light off shall be considered anomalous. In the CATS dataset, users can choose (or not) to use the available context, and external stimuli, to test the usefulness of the context for detecting anomalies in this simulation.

Pure signal ideal for robustness-to-noise analysis. The simulated signals are provided without noise: while this may seem unrealistic at first, it is an advantage since users of the dataset can decide to add on top of the provided series any type of noise and choose an amplitude. This makes it well suited to test how sensitive and robust detection algorithms are against various levels of noise.

No missing data. You can drop whatever data you want to assess the impact of missing values on your detector with respect to a clean baseline.

Change Log

Version 2

Metadata: we include a metadata.csv with information about:

Anomaly categories

Root cause channel (signal in which the anomaly is first visible)

Affected channel (signal in which the anomaly might propagate) through coupled system dynamics

Removal of anomaly overlaps: version 1 contained anomalies which overlapped with each other resulting in only 190 distinct anomalous segments. Now, there are no more anomaly overlaps.

Two data files: CSV and parquet for convenience.

[1] Example Benchmark of Anomaly Detection in Time Series: “Sebastian Schmidl, Phillip Wenig, and Thorsten Papenbrock. Anomaly Detection in Time Series: A Comprehensive Evaluation. PVLDB, 15(9): 1779 - 1797, 2022. doi:10.14778/3538598.3538602”

About Solenix

Solenix is an international company providing software engineering, consulting services and software products for the space market. Solenix is a dynamic company that brings innovative technologies and concepts to the aerospace market, keeping up to date with technical advancements and actively promoting spin-in and spin-out technology activities. We combine modern solutions which complement conventional practices. We aspire to achieve maximum customer satisfaction by fostering collaboration, constructivism, and flexibility.

Time Series PILE

The Time-series Pile is a large collection of publicly available data from diverse domains, ranging from healthcare to engineering and finance. It comprises of over 5 public time-series databases, from several diverse domains for time series foundation model pre-training and evaluation.

  Time Series PILE Description

We compiled a large collection of publicly available datasets from diverse domains into the Time Series Pile. It has 13 unique domains of data… See the full description on the dataset page: https://huggingface.co/datasets/AutonLab/Timeseries-PILE.

Network traffic datasets created by Single Flow Time Series Analysis

Datasets were created for the paper: Network Traffic Classification based on Single Flow Time Series Analysis -- Josef Koumar, Karel Hynek, Tomáš Čejka -- which was published at The 19th International Conference on Network and Service Management (CNSM) 2023. Please cite usage of our datasets as:

J. Koumar, K. Hynek and T. Čejka, "Network Traffic Classification Based on Single Flow Time Series Analysis," 2023 19th International Conference on Network and Service Management (CNSM), Niagara Falls, ON, Canada, 2023, pp. 1-7, doi: 10.23919/CNSM59352.2023.10327876.

This Zenodo repository contains 23 datasets created from 15 well-known published datasets which are cited in the table below. Each dataset contains 69 features created by Time Series Analysis of Single Flow Time Series. The detailed description of features from datasets is in the file: feature_description.pdf

In the following table is a description of each dataset file:

This data is synced hourly from https://github.com/CSSEGISandData/COVID-19. All credit is to them.

Latest Confirmed Cases

@(https://data.world/shad/covid-analysis/workspace/query?datasetid=covid-19-time-series-data&queryid=e066701e-fa8d-4c9f-97f8-aab3a6f219a8)

​

I have also added confirmed_pivot.csv which gives a slightly more workable view of the data. Extra columns/day makes things difficult.

@(https://data.world/shad/covid-analysis/workspace/file?datasetid=covid-19-time-series-data&filename=confirmed_pivot)

​

#

Identifying change points and/or anomalies in dynamic network structures has become increasingly popular across various domains, from neuroscience to telecommunication to finance. One particular objective of anomaly detection from a neuroscience perspective is the reconstruction of the dynamic manner of brain region interactions. However, most statistical methods for detecting anomalies have the following unrealistic limitation for brain studies and beyond: that is, network snapshots at different time points are assumed to be independent. To circumvent this limitation, we propose a distribution-free framework for anomaly detection in dynamic networks. First, we present each network snapshot of the data as a linear object and find its respective univariate characterization via local and global network topological summaries. Second, we adopt a change point detection method for (weakly) dependent time series based on efficient scores, and enhance the finite sample properties of change point method by approximating the asymptotic distribution of the test statistic using the sieve bootstrap. We apply our method to simulated and to real data, particularly, two functional magnetic resonance imaging (fMRI) datasets and the Enron communication graph. We find that our new method delivers impressively accurate and realistic results in terms of identifying locations of true change points compared to the results reported by competing approaches. The new method promises to offer a deeper insight into the large-scale characterizations and functional dynamics of the brain and, more generally, into the intrinsic structure of complex dynamic networks. Supplemental materials for this article are available online.

We include Stata syntax (dummy_dataset_create.do) that creates a panel dataset for negative binomial time series regression analyses, as described in our paper "Examining methodology to identify patterns of consulting in primary care for different groups of patients before a diagnosis of cancer: an exemplar applied to oesophagogastric cancer". We also include a sample dataset for clarity (dummy_dataset.dta), and a sample of that data in a spreadsheet (Appendix 2).

The variables contained therein are defined as follows:

case: binary variable for case or control status (takes a value of 0 for controls and 1 for cases).

patid: a unique patient identifier.

time_period: A count variable denoting the time period. In this example, 0 denotes 10 months before diagnosis with cancer, and 9 denotes the month of diagnosis with cancer,

ncons: number of consultations per month.

period0 to period9: 10 unique inflection point variables (one for each month before diagnosis). These are used to test which aggregation period includes the inflection point.

burden: binary variable denoting membership of one of two multimorbidity burden groups.

We also include two Stata do-files for analysing the consultation rate, stratified by burden group, using the Maximum likelihood method (1_menbregpaper.do and 2_menbregpaper_bs.do).

Note: In this example, for demonstration purposes we create a dataset for 10 months leading up to diagnosis. In the paper, we analyse 24 months before diagnosis. Here, we study consultation rates over time, but the method could be used to study any countable event, such as number of prescriptions.

Over time statistical outputs (and time series data) may be subject to revisions or corrections. Revisions are generally planned, and are the result of either improvements in statistical methods or the availability of additional data. For example, the annual mid-year population estimates are revised after a census to take account of the additional information gained from the census results. Details of planned revisions are held within the Metadata alongside each publication. Corrections are unplanned and occur when errors in either the statistical data or methodology are found after release of the data. The latest correction to these datasets was in September 2018, for more information please see the revisions and corrections page. This time series section provides access to the latest time series data, taking into account any revisions or corrections over the years. Note: Tables are mainly offered for the purposes of extracting figures. Due to the size of some of the sheets they are not recommended for printing.

BayModTS is a FAIR workflow for processing highly variable and sparse data. The code and results of the examples in the BayModTS paper are stored in this repository. A maintained version of BayModTS that can be applied to your personal applications can be found on Git Hub.

Time Series Analytics Market Outlook

As per our latest research, the global Time Series Analytics market size reached USD 6.4 billion in 2024, demonstrating robust momentum driven by rapid digital transformation and the proliferation of IoT and connected devices. The market is projected to expand at a CAGR of 13.2% from 2025 to 2033, reaching an estimated USD 19.1 billion by 2033. This impressive growth is propelled by increasing demand for advanced analytics solutions across industries, particularly as organizations seek to harness real-time data for improved decision-making and operational efficiency.

One of the primary growth factors for the Time Series Analytics market is the exponential rise in data generated from connected devices and sensors, especially within sectors such as manufacturing, energy, and utilities. As organizations embrace Industry 4.0, the need to analyze vast streams of time-stamped data for operational insights, predictive maintenance, and process optimization becomes paramount. The adoption of advanced analytics platforms capable of handling high-velocity, high-volume time series data empowers enterprises to identify patterns, forecast trends, and detect anomalies in real-time, thus driving business value and competitive advantage.

Another significant driver is the growing integration of artificial intelligence (AI) and machine learning (ML) algorithms within time series analytics solutions. These technologies enhance the accuracy and scalability of forecasting, anomaly detection, and risk management applications, enabling organizations to proactively mitigate risks and capitalize on emerging opportunities. The increasing focus on data-driven decision-making across sectors such as BFSI, healthcare, and retail further accelerates the adoption of time series analytics, as businesses seek to optimize customer experiences, streamline operations, and ensure regulatory compliance through actionable insights derived from real-time data streams.

The surge in cloud adoption is also reshaping the Time Series Analytics market, with organizations favoring cloud-based analytics platforms for their scalability, flexibility, and cost-effectiveness. Cloud deployment models facilitate seamless integration with diverse data sources, support remote collaboration, and provide access to advanced analytics capabilities without the overhead of on-premises infrastructure. This shift is particularly pronounced among small and medium enterprises (SMEs) and emerging markets, where cloud solutions lower entry barriers and democratize access to sophisticated analytics tools. As cloud-native, AI-driven analytics platforms become increasingly prevalent, the market is poised for sustained growth and innovation over the forecast period.

Regionally, North America continues to dominate the Time Series Analytics market, accounting for the largest share in 2024, followed by Europe and the Asia Pacific. The presence of leading technology vendors, high digital maturity, and early adoption of advanced analytics solutions underpin North America's leadership. Meanwhile, Asia Pacific is witnessing the fastest growth, fueled by rapid industrialization, digital transformation initiatives, and expanding investments in IoT and smart technologies. As organizations across regions accelerate their digital journeys and prioritize data-driven strategies, the global time series analytics landscape is set for dynamic expansion through 2033.

Component Analysis

The Component segment of the Time Series Analytics market is bifurcated into software and services, each playing a pivotal role in driving market growth and adoption. Software solutions form the backbone of the market, encompassing time series databases, analytics platforms, and visualization tools that enable enterprises to ingest, process, and analyze vast volumes of temporal data. The increasing sophistication of these software offerings, including support for AI and ML-driven analytics, real-time processing, and seamless inte

Version 1.0 - This version is the final revised one.

This is the LamaH-CE dataset accompanying the paper: Klingler et al., LamaH-CE | LArge-SaMple DAta for Hydrology and Environmental Sciences for Central Europe, published at Earth System Science Data (ESSD), 2021 (https://doi.org/10.5194/essd-13-4529-2021).

LamaH-CE contains a collection of runoff and meteorological time series as well as various (catchment) attributes for 859 gauged basins. The hydrometeorological time series are provided with daily and hourly time resolution including quality flags. All meteorological and the majority of runoff time series cover a span of over 35 years, which enables long-term analyses with high temporal resolution. LamaH is in its basics quite sililar to the well-known CAMELS datasets for the contiguous United States (https://doi.org/10.5194/hess-21-5293-2017), Chile (https://doi.org/10.5194/hess-22-5817-2018), Brazil (https://doi.org/10.5194/essd-12-2075-2020), Great Britain (https://doi.org/10.5194/essd-12-2459-2020) and Australia (https://doi.org/10.5194/essd-13-3847-2021), but new features like additional basin delineations (intermediate catchments) and attributes allow to consider the hydrological network and river topology in further applications.

We provide two different files to download: 1) Hydrometeorological time series with daily and hourly resolution, which requires decompressed about 70 GB of free disk space. 2) Hydrometeorological time series only with daily resolution, which requires 5 GB. Beyond the temporal resolution of the time series, there are no differences.

Note: It is recommended to read the supplementary info file before using the dataset. For example, it clarifies the time conventions and that NAs are indicated by the number -999 in the runoff time series.

Disclaimer: We have created LamaH with care and checked the outputs for plausibility. By downloading the dataset, you agree that we nor the provider of the used source datasets (e.g. runoff time series) cannot be liable for the data provided. The runoff time series of the German federal states Bavaria and Baden-Württemberg are retrospective checked and updated by the hydrographic services. Therefore, it might be appropriate to obtain more up-to-date runoff data from Bavaria (https://www.gkd.bayern.de/en/rivers/discharge/tables) and Baden-Württemberg (https://udo.lubw.baden-wuerttemberg.de/public/p/pegel_messwerte_leer). Runoff data from the Czech Republic may not be used to set up operational warning systems (https://www.chmi.cz/files/portal/docs/hydro/denni_data/Podminky_uziti.pdf).

License: This work is licensed with CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/). This means that you may freely use and modify the data (even for commercial purposes). But you have to give appropriate credit (associated ESSD paper, version of dataset and all sources which are declared in the folder "Info"), indicate if and what changes were made and distribute your work under the same public license as the original.

Additional references: We ask kindly for compliance in citing the following references when using LamaH, as an agreement to cite was usually a condition of sharing the data: BAFU (2020), CHMI (2020), GKD (2020), HZB (2020), LUBW (2020), BMLFUW (2013), Broxton et al. (2014), CORINE (2012), EEA (2019), ESDB (2004), Farr et al. (2007), Friedl and Sulla-Menashe (2019), Gleeson et al. (2014), HAO (2007), Hartmann and Moosdorf (2012), Hiederer (2013a, b), Linke et al. (2019), Muñoz Sabater et al. (2021), Muñoz Sabater (2019a), Myneni et al. (2015), Pelletier et al. (2016), Toth et al. (2017), Trabucco and Zomer (2019), and Vermote (2015). These references are listed in detail in the accompanying paper.

Supplements: We have created additional files after publication (therefore non peer-reviewed): 1) Shapefiles for reservoirs (points) and cross-basin water transfers (lines) including several attributes as well as tables with information about the accumulated storage volume and effective catchment area (considerung artificial in- and outflows) for every runoff gauge. 2) Water quality data (e.g. dissolved oxygen, water temperature, conductivity, NO3-N), which are suitable to the gauges. The data for water quality may not be used for commercial purposes. If you are interessted, just send us an email with your name, affiliation and the intended purpose for the requested files to the address listed below. If you find any errors in the dataset, feel free to send us an email to: christoph.klingler@boku.ac.at

Cross-validation biases time series model selection even when used with VARs or with models that have martingale-like errors.

The mid-year estimates refer to the population on 30 June of the reference year and are produced in line with the standard United Nations (UN) definition for population estimates. They are the official set of population estimates for the UK and its constituent countries, the regions and counties of England, and local authorities and their equivalents.

THE MISSION

The story behind the dataset is how to apply LSTM architecture to understand and apply multiple variables together to contribute more accuracy towards forecasting.

THE CONTENT

Air Pollution Forecasting The Air Quality dataset.

This is a dataset that reports on the weather and the level of pollution each hour for five years at the US embassy in Beijing, China.

The data includes the date-time, the pollution called PM2.5 concentration, and the weather information including dew point, temperature, pressure, wind direction, wind speed and the cumulative number of hours of snow and rain. The complete feature list in the raw data is as follows:

No: row number year: year of data in this row month: month of data in this row day: day of data in this row hour: hour of data in this row pm2.5: PM2.5 concentration DEWP: Dew Point TEMP: Temperature PRES: Pressure cbwd: Combined wind direction Iws: Cumulated wind speed Is: Cumulated hours of snow Ir: Cumulated hours of rain We can use this data and frame a forecasting problem where, given the weather conditions and pollution for prior hours, we forecast the pollution at the next hour.